Do we really gesture more when it is more difficult?

Representational co-speech gestures are generally assumed to be increasingly produced in more difficult compared with easier verbal tasks, as maintained by theories suggesting that gestures arise from processing difficulties during speech production. However, the gestures-as-simulated-action framework proposes that more representational gestures are produced with stronger rather than weaker mental representations that are activated in terms of mental simulation in the embodied cognition framework. We tested these two conflicting assumptions by examining verbal route descriptions that were accompanied by spontaneous directional gestures. Easy descriptions with strong activation were accompanied more often by gestures than difficult descriptions with weak activation. Furthermore, only gesture-speech matches-but not gesture-picture matches-were increasingly produced with difficult lateral directions compared with easy nonlateral directions. We argue that lateral gesture-speech matches underlie stronger activated mental representations in mental imagery. Thus, all results are in line with the gestures-as-simulated-action framework and provide evidence against the view that gestures result from processing difficulties.